Type:

Year:

Supervisor:

Institution:

Page(s):

Abstract:

The design of system architectures is a complex task that involves major challenges. During this activity, system designers must create design alternatives and compare them with each other in order to select the most appropriate architecture according to a set of criteria. In order to study different alternatives, designers usually have to limit their comparative study to a small part of the design space, which may be composed of a huge number of solutions. Traditionally, the architectural design process is mainly driven by the judgment and experience of designers, and the alternatives selected are adapted versions of known solutions. The risk is therefore to select a relevant but sub-optimal solution. To gain confidence in the optimality of the selected solution, the coverage of the design space must be increased. The use of computational architecture synthesis methods has proven to be an efficient way to support designers in the design of engineering artifacts (structures, electrical circuits...). To assist system designers during the architecture design process, a computational method for complex systems is defined. This method uses an evolutionary approach (genetic algorithms) to guide the process of exploring the design space towards optimal areas. The initial population of the genetic algorithm is created using a computational architecture synthesis technique that allows the creation of different physical architectures and allocation tables for a given functional architecture. The method provides optimal solutions to the design problem posed. These solutions can then be used by designers for more detailed comparative studies or for negotiations with system suppliers.

Keywords:

system architecture, combinatorial optimization, synthesis of alternatives